Numpy dependency causing errors on linux
Closed this issue · 4 comments
My student and I ran into some issues installing TheJoker on our HPC cluster. When installing on Python 3.10 or above, we get the following output
Collecting thejoker
Using cached thejoker-1.3.0.tar.gz (2.1 MB)
Installing build dependencies ... done
Getting requirements to build wheel ... error
error: subprocess-exited-with-error
× Getting requirements to build wheel did not run successfully.
│ exit code: 1
╰─> [20 lines of output]
Traceback (most recent call last):
File "/home/douglste-laf/.conda/envs/newjoker/lib/python3.10/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 353, in <module>
main()
File "/home/douglste-laf/.conda/envs/newjoker/lib/python3.10/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 335, in main
json_out['return_val'] = hook(**hook_input['kwargs'])
File "/home/douglste-laf/.conda/envs/newjoker/lib/python3.10/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py", line 118, in get_requires_for_build_wheel
return hook(config_settings)
File "/tmp/pip-build-env-kot7ogin/overlay/lib/python3.10/site-packages/setuptools/build_meta.py", line 327, in get_requires_for_build_wheel
return self._get_build_requires(config_settings, requirements=[])
File "/tmp/pip-build-env-kot7ogin/overlay/lib/python3.10/site-packages/setuptools/build_meta.py", line 297, in _get_build_requires
self.run_setup()
File "/tmp/pip-build-env-kot7ogin/overlay/lib/python3.10/site-packages/setuptools/build_meta.py", line 313, in run_setup
exec(code, locals())
File "<string>", line 11, in <module>
File "/tmp/pip-build-env-kot7ogin/overlay/lib/python3.10/site-packages/twobody/__init__.py", line 28, in <module>
from .anomaly import *
File "/tmp/pip-build-env-kot7ogin/overlay/lib/python3.10/site-packages/twobody/anomaly.py", line 9, in <module>
from .wrap import (cy_mean_anomaly_from_eccentric_anomaly,
File "twobody/wrap.pyx", line 1, in init twobody.wrap
ValueError: numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject
[end of output]
note: This error originates from a subprocess, and is likely not a problem with pip.
error: subprocess-exited-with-error
× Getting requirements to build wheel did not run successfully.
│ exit code: 1
╰─> See above for output.
note: This error originates from a subprocess, and is likely not a problem with pip.
It looks like this is a Numpy error, though they also blame pip. Resolving it requires using an older version of Numpy, none of which are compatible with Python>=3.10
I was able to solve this by requiring python=3.9 and numpy==1.26.4 when I created my conda environment. This is the full set of commands I used to get a working installation
conda create -n thejoker python=3.9 scipy==1.10 numpy==1.26.4
conda activate thejoker
conda install -c anaconda gcc
conda install h5py netCDF4
pip install thejoker
The cluster is running Rocky Linux (I believe Version 8) on an x86 64bit architecture. Unfortunately the documentation is a bit out of date at the moment, so I can't be 100% sure, but I've seen other packages running into the same error.
Sorry for the slowness and thanks for reporting! I just released a new version that I think should resolve the numpy compatibility issues related to their 2.0 release. You have a working install, so I understand if you don't want to test, but if you feel like making a new environment without the constraints above, can you see if it all works now?
I was successfully able to install thejoker with no errors, but it installed numpy=1.26.4 along the way, instead of numpy=2.1.0. When I tried installing in an environment that already had numpy=2.1.0, in uninstalled numpy and reverted it to the older version.
OK good -- thanks for checking!
Yes unfortunately this is expected: the current need for numpy<2 comes from pytensor (core dep of pymc), which requires numpy<2. But now things are set up so when that when pytensor is updated, thejoker should work seamlessly on numpy>=2.0.
I'm going to close this for now, but if anything else comes up feel free to re-open or make a new issue. Thanks again for reporting!